Develop

See merge request !6

1) Fully symmetric system ala:
  |A   B^T| |x| = |f|
  |B       0| |y|    |g|
2) Still have positive-definite system when stabilized in almost anti-sym formulation because:
  => d + f1 + f2 + h > 0 (quadratic term)

- Only affect R13 inflow test case (adjusted referrors, qualitative convergence remains)
- System is now fully antisymmetric

- density formulation has no prob
- but pressure formulation is then non (anti-)symmetric
- therefore use mass balance to eliminate div(u)=f_mass from LHS of energy equation to remain (a-)sym
- This needs the removal of unneeded mass source from heat testcases
- Also needs adapting R13 cases such that for mass source, the same energy source is applied s.t. energy equation remains equal
  - The latter is actually only needed to compare with WK19 where div(u) was inconsistently neglected

- Replace un-unw = eps chi ((p-pw)+sigmann)
- By un-unw = eps chi ((p-pw)-sigmann)
- Should be allowed because kappa1 is arbitrary following Torrilhon2017 (see sec 5.2)
- Resulting system is then anti-symm
- But can be made fully sym by partial row inversion
- Add esol for change inflowBC but not used for now

- This replaces the pressure-driven example
- Matches better with Torrilhon2016 that also uses a constant dimensionless
force
- The plots, however, are equal because it's completely equivalent

- Does not work to replace x[0] in function call

- Needs slightly modification of referrors, but trend stays the same

- Just reordering

- Try to fix blocking CI with manual

Update to version 1.1

See merge request !5